Explore cutting-edge research in embodied intelligence through this seminar featuring presentations on various topics. Delve into interactive robot training for non-Markov tasks, long-horizon manipulation of unknown objects, task-informed abstractions, and the geometry of contextualized word representations. Learn about hybrid inference methods, advances in SLAM, causal navigation using neural networks, and cooperative learning for machine reading comprehension. Discover object SLAM optimization techniques, a platform for massively multiagent research, and implicit representations of meaning in language models. Gain insights into the latest developments in AI, robotics, natural language processing, and machine learning from MIT researchers.
Overview
Syllabus
Introduction
Demonstration
Framework
Mom
Video Damage
Learning Tax Informed Instructions
Conclusion
Word Representations
Causal Navigation
Question Answering
Consensus Informed Optimization
Neural Mmo
Taught by
MIT Embodied Intelligence
